Attention is currently required from: Nico Huber, Tim Wawrzynczak.
5 comments:
File src/soc/intel/common/block/gpio/gpio.c:
Patch Set #9, Line 214: 120 / (sizeof(uint32_t) * BITS_PER_BYTE) + 1
DIV_ROUND_UP(120, sizeof(uint32_t) * BITS_PER_BYTE)
Patch Set #9, Line 218: size_t bit_offset = irq % 32;
assert(word_offset < ARRAY_SIZE(gpio_ioapic_irqs_used)) or just return after printing an error message?
Patch Set #9, Line 225: size_t bit_offset = irq % 32;
assert(word_offset < ARRAY_SIZE(gpio_ioapic_irqs_used)) or just return false after printing an error message?
Patch Set #9, Line 263: set_ioapic_used(irq);
This should be done before line 236, right? If a platform does not select SOC_INTEL_COMMON_BLOCK_GPIO_ITSS_POL_CFG, then this will never get called. You might have to change the order getween PAD_CFG0_ROUTE_IOAPIC and CONFIG(SOC_INTEL_COMMON_BLOCK_GPIO_ITSS_POL_CFG) checks.
Not for this change, but I think SOC_INTEL_COMMON_BLOCK_GPIO_ITSS_POL_CFG is selected for all platforms using GPIO common block driver. Probably we should drop that config altogether since all Intel SoCs seem to have this support.
File src/soc/intel/common/block/include/intelblocks/gpio.h:
Patch Set #9, Line 232: routed to IOAPIC
I know this is obvious from the function implementation, but should we add a comment here indicating that it checks if any GPIO is currently routed to IOAPIC or GPIO is routed to IOAPIC at the time this call is made? Or do you think that is just adding noise?
To view, visit change 49407. To unsubscribe, or for help writing mail filters, visit settings.